Sex dependency and genetic modulation of emotional processing and memory: a behavioural and imaging study
A cumulative dissertation
submitted to the Faculty of Psychology, University of Basel,
in partial fulfilment of the requirements for the degree of Doctor of philosophy
by
M.Sc. Klara Spalek from Thun, Switzerland
Basel, Switzerland June 2014
The original document saved on the document server of the University of Basel edoc.unibas.ch
This work is licenced under the contract „Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Switzerland“ (CC BY-NC-ND 3.0 CH). The complete licence can be found under creativecommons.org/licenses/by-nc-nd/3.0/ch/ .
Approved by the Faculty of Psychology at the request of
Professor Dr. med. Andreas Papassotiropoulos Professor Dr. med. Dominique J.-F. de Quervain
Basel, the 2. July 2014
Attribution-NonCommercial-NoDerivs 3.0 Switzerland (CC BY-NC-ND 3.0 CH)
You are free to: Share — copy and redistribute the material in any medium or format The licensor cannot revoke these freedoms as long as you follow the license terms.
Under the following terms:
Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any
reasonable manner, but not in any way that suggests the licensor endorses you or your use.
NonCommercial — You may not use the material for commercial purposes.
NoDerivatives — If you remix, transform, or build upon the material, you may not distribute the modified material.
No additional restrictions — You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.
Notices:
You do not have to comply with the license for elements of the material in the public domain or where your use is permitted by an applicable exception or limitation.
No warranties are given. The license may not give you all of the permissions necessary for your intended use. For example, other rights such as publicity, privacy, or moral rights may limit how you use the material.
ABSTRACT
1 ABSTRACT
Emotional processing and episodic memory are topics of great interest in neuroscience research. It is known that both of these two cognitive processes can be influenced by a variety of factors. The scope of this thesis is to highlight the importance and describe the role of two of these factors, namely sex and genetics. Our investigation on this topic consists of results that have been reported in five peer-reviewed publications, where methods from different disciplines like neuroimaging, psychoneuroendocrinology, and epigenetics were combined.
In order to assess sex-dependent differences in emotional processing and episodic memory, we conducted (in our first publication) behavioural and imaging analyses within a large sample of healthy young subjects. Our results point to differences between the sexes in emotional appraisal as well as setting-dependent differences in memory performance of pictorial information, which seem to be independent of each other. We additionally investigated the modulatory character of endogenous testosterone levels on emotional processing and memory (in the second publication). The results may suggest a role of testosterone in enhancing memory performance for neutral stimuli by increasing the biological salience of this information, as indicated by increased arousal ratings and amygdala reactivity to these stimuli.
To further investigate the genetic modulation of emotional processing and episodic memory we focused on the role of three genes (neurotrophic tyrosine kinase receptor type 2 (NTRK2), protein kinase C alpha (PRKCA) and brain derived neurotrophic factor (BDNF)).
We provided (in out third publication) evidence for a role of a NTRK2 variant in the emotion processing of positive stimuli in healthy young subjects and additionally found NTRK2- dependent differences in white matter measures as well as methylation levels. In reference to episodic memory, we found (in the fourth publication) that PRKCA plays a role in memory performance of healthy subjects and is also associated with specific symptoms of posttraumatic stress disorder (PTSD) as well as the risk to develop PTSD in genocide survivors. Finally (in the fifth publication), by combining original data with meta-analytic techniques, we found no association between a genetic variant of the BDNF gene and hippocampal volumes in our original data and show that the weak association in the meta- analysis is moderated by measuring techniques, publication year and sample size.
Taken together, these results support the presence of sex-dependent differences in emotional processing as well as episodic memory and emphasize the role of specific genes in these processes.
TABLE OF CONTENTS
TABLE OF CONTENTS
FIGURE INDEX ... 4
ACKNOWLEDGMENTS ... 5
ABBREVIATIONS ... 6
1. INTRODUCTION ... 8
2. THEORETICAL BACKGROUND ... 11
2.1. EMOTION ... 11
2.1.1. EMOTION PROCESSES ... 11
2.1.2. NEURONAL BASIS OF EMOTION ... 12
2.2. MEMORY ... 13
2.2.1. MEMORY SYSTEMS ... 13
2.2.2. NEURONAL BASIS OF MEMORY ... 15
2.2.3. MOLECULAR BASIS OF MEMORY ... 17
2.3. MEMORYMODULATIONTHROUGHEMOTION ... 19
2.4. MODULATINGFACTORSOFEMOTIONALPROCESSINGANDEPISODIC MEMORY ... 21
2.4.1. SEX ... 21
2.4.2. GENETICS ... 23
3. METHODS ... 25
3.1. NEUROIMAGING ... 25
3.2. PSYCHONEUROENDOCRINOLOGY ... 26
3.3. GENETICANALYSIS ... 27
3.4. EPIGENETICANALYSIS ... 28
4. ORIGINAL RESEARCH PAPERS ... 29
4.1. GENDER-DEPENDENTDISSOCIATIONBETWEENEMOTIONALAPPRAISAL ANDMEMORY:ALARGE–SCALEBEHAVIOURALANDFMRISTUDY ... 29
4.2. TESTOSTERONELEVELSINHEALTHYMENARERELATEDTO AMYGDALAREACTIVITYANDMEMORYPERFORMANCE ... 71
TABLE OF CONTENTS
3 4.3. GENETICVARIANTSOFNTRK2AREASSOCIATEDWITHEMOTION
PROCESSING,AWHITE-MATTERMEASUREANDDNAMETHYLATIONLEVELS
INHEALTHYYOUNGSUBJECTS ... 103
4.4. PKC!ISGENETICALLYLINKEDTOMEMORYCAPACITYINHEALTHY SUBJECTSANDTORISKFORPOSTTRAUMATICSTRESSDISORDERIN GENOCIDESURVIVORS ... 140
4.5. THEASSOCIATIONOFTHEBDNFVAL66METPOLYMORPHISMANDTHE HIPPOCAMPALVOLUMESINHEALTHYHUMANS:AJOINTMETA-ANALYSISOF PUBLISHEDANDNEWDATA ... 177
5. DISCUSSION ... 190
6. REFERENCES ... 197
DECLARATION BY CANDIDATE ... 212
FIGURE INDEX
FIGURE INDEX
Figure 1. Brain activations during emotion perception and processing.
page 13
Figure 2. Overview of memory systems.
page 15
Figure 3. Molecular mechanisms of short-term and long-term memory.
page 18
Figure 4. Schematic representation of the components of the emotional memory enhancement effect.
page 20
Figure 5. Sex-dependent brain activation differences during emotion processing of negative and positive emotions.
page 22
ACKNOWLEDGMENTS
5 ACKNOWLEDGMENTS
!
In my time as a PhD student in the lab of Professor Dominique de Quervain and Professor Andreas Papassotiropoulos I witnessed and was involved in neuroscience research of excellent quality. Since the lab’s members are coming from different disciplines like molecular biology, psychology, medicine, physics and bio-informatics I had the unique and extremely valuable opportunity to acquire some knowledge from all these fields. It is an honour for me to have completed my PhD in such an outstanding research group.
I would therefore like to thank my supervisors Professor Dominique de Quervain and Professor Andreas Papassotiropoulos for their support throughout these years. Their expert advice, guidance, personal encouragement and understanding provided me with invaluable knowledge, skill and confidence for my future work in science.
I would also like to thank all my working colleagues, who supported me during these years in terms of teaching me their knowledge as well as giving me psychologically support. Some of them deserve particular mention: M.Sc. Matthias Fastenrath and Dr. David Coynel for their help with imaging analyses. Dr. Angela Heck for her introduction to genetic analyses. Finally, Dr. Vanja Vukojevic for providing me with knowledge in the field of epigenetics. The friendship with M.Sc. Matthias Fastenrath and M.Sc. Nathalie Schicktanz evolving within my PhD years was a great support.
Last but not least, I would specially like to thank my partner Dr. Manolis Sifalakis for his enormous support and patience in these years and to my parents, Hana and Jaroslav Spalek, for their support, which gave me endurance in the processes to get my PhD.
ABBREVIATIONS
ABBREVIATIONS
!
AC: adenylyl cyclase
ACC: anterior cingulate cortex AF: activating factor
AMPA: "-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid ATP: adenosine triphosphate
BDNF: brain derived neurotrophic factor BLA: basolateral amygdala
BOLD: blood oxygenation level-dependent C/EBP: CCAAT-box-enhancer binding protein Ca2+: calcium
CaMK II: Ca2+-calmodulin-dependent kinase II cAMP: cyclic adenosine monophosphate CpG: cytosine-guanine dinucleotide CRE: cAMP response element
CREB-1: cAMP response element binding protein-1 CREB-2: cAMP response element binding protein-2 CSF: cerebrospinal fluid
DLPFC: dorsolateral prefrontal cortex DNA: deoxyribonucleic acid DTI: diffusion tensor imaging DWI: diffusion weighted imaging EF1": elongation factor 1"
EMG: electromyography EPI: echo-planar imaging ERP: event-related potential FA: fractional anisotropy
fMRI: functional magnetic resonance imaging FSL: fMRI of the brain software library GSA: gene set analysis
GWAS: genome-wide association study HRF: hemodynamic response function IAPS: international affective picture system
ABBREVIATIONS
7 ICV: intracranial volume
K+: potassium
LTP: long-term potentiation
MAPK: mitogen-activated protein kinase MD: mean diffusivity
mFC: medial frontal cortex mPFC: medial prefrontal cortex MRI: magnetic resonance imaging MTL: medial temporal lobe
NAA: network-assisted analysis NE: norepinephrine
NMDA: N-Methyl-D-Aspartate NO: nitric oxide
NTRK2: neurotrophic tyrosine kinase receptor type 2 OFC: orbitofrontal cortex
PBA: pathway-based analysis PET: positron emission tomography PFC: prefrontal cortex
PKA: cAMP-dependent protein kinase A PKC: protein kinase C
PKC": protein kinase C alpha PRKCA: protein kinase C alpha PTSD: posttraumatic stress disorder TRKB: tyrosine kinase receptor B vmPFC: ventromedial prefrontal cortex
INTRODUCTION
1. INTRODUCTION
Our daily life consists of perception and processing of information, and reactions in response to this information. Emotion and memory play a central role in these processes. A good portion of the information, we are daily confronted with, are of emotional content or in some cases we simply assign emotional meanings to this information. Emotions can as well describe a state we currently might be in; for example we can feel happy having past an exam or we can feel angry after quarrel. Furthermore, in patients with different psychiatric disorders the main characteristic is a dysregulation of emotions (Cole, Michel, & Teti, 1994; Kring &
Sloan, 2009). Thus, it is obvious, that emotions play an essential role in our lives and it is not surprising that emotion processes including perception, processing and response are since long time a topic of great interest in research. The first theories about emotions date back to 1890 to the James-Lang-Theory. Today, we have developed a much more elaborated and extended concept about emotional perception, processing and reactions by connecting the knowledge from different research disciplines like psychology, molecular biology, genetics, epigenetics and neuroimaging, thanks to their diversity of methodologies. It is known that different neurotransmitters like serotonin, noradrenaline, adrenaline and dopamine are involved in emotional processes (Bear, Connors, & Paradiso, 2009; Lövheim, 2012; McGeer
& McGeer, 1980). Alongside brain activations and structural characteristics especially in the amygdala, as well as in several other brain regions like prefrontal cortex (PFC), orbitofrontal cortex (OFC), anterior cingulate cortex (ACC), insula and hypothalamus are playing a central role in emotions (Dalgleish, 2004; Phan, Wager, Taylor, & Liberzon, 2004). In this thesis we focus specifically on the emotion processing stage.
On the other hand, memory, which is the ability to encode, store and retrieve information and experiences, is equally essential for our daily functioning. The best proof of the importance of our ability to form memories comes from cases where memory functions are not intact anymore like in the famous patient H.M. or in patients suffering from the Alzheimer disease or Amnesia (the loss of memory for a specific period or loss of the ability to acquire new memories). The case of H.M. as well as studies on animals and amnestic patients further lead to the differentiation of memory systems like declarative and non- declarative memory with its subsystems (Squire 2004). The work of this thesis focuses mainly on episodic memory, a sub-system of declarative memory. A major breakthrough, for the understanding of the molecular basis of memory, was Eric Kandel’s work with the sea snail Aplysia Californica for which he received in 2000 the Nobel price for physiology and
INTRODUCTION
9 medicine. On the neuronal level, research has provided evidence for the central role of the hippocampus in episodic memory (Squire & Wixted, 2011), as well as other brain regions like the cortex areas, which are initially involved in the processing of the information that is later remembered (Squire & Kandel, 2009). An important modulating factor of episodic memory performance is the perceived emotionality of the material (Roozendaal & McGaugh, 2011).
Namely, information of emotionally arousing content is supposed to lead to a more elaborated processing and thus better remembering due to its salience (LaBar & Cabeza, 2006). This mechanism is evolutionary driven, since it is useful to remember threatening or rewarding situations for future behaviour. This emotional enhancement effect is partially mediated through amygdala activity (Cahill et al., 1996; McGaugh, 2004; McGaugh & Roozendaal, 2002).
Both, emotion processing and episodic memory can be influenced by several factors amongst others by sex and genetics, which are the focus of this thesis. Sex-dependent differences in emotional processing and episodic memory have been already reported in literature (emotion processing: Bradley et al., 2001; Gard & Kring, 2007; Lithari et al., 2010;
episodic memory: Andreano & Cahill, 2009; Bloise & Johnson, 2007; Herlitz et al., 1997;
Herlitz et al., 2013; de Frias et al., 2006). Specifically, women process especially negative material more intensively and perform in general better than men on episodic memory tasks.
Concerning the genetic modulation of emotion processing and episodic memory, substantial amount of research provides evidence for an association between genes or gene clusters and these behavioural traits (Bevilacqua & Goldman, 2011; Papassotiropoulos & de Quervain, 2011). The on-going methodological progresses like the introduction of genome-wide association studies (GWAS) or pathway analysis in this discipline enable researchers to develop an understanding of the role of genes in these processes.
The aim of the present doctoral thesis consists in the contribution to a better understanding of the role of sex and genetics in emotional processing and emotional episodic memory. The intent has been to develop a more elaborated picture about these processes by combining insights from different areas (neuroimaging, psychoneuroendocrinology, genetics, and epigenetics) in investigating them. We envision that these results might provide information about the mechanisms of these processes in healthy young subjects and might add useful insights about the dysfunctions in these processes in psychiatric disorders.
The findings of this thesis have been reported in the following five publications:
INTRODUCTION
1. Gender-dependent dissociation between emotional appraisal and memory: a large–scale behavioural and fMRI study.
Spalek, K., Fastenrath, M., Ackermann, S., Auschra, B., Coynel, D., Frey, J., Gschwind, L., Hartmann, F., van der Maarel, N., Papassotiropoulos, A., de Quervain, D. J.-F., &
Milnik A., in preparation (a).
2. Testosterone levels in healthy men are related to amygdala reactivity and memory performance.
Ackermann, S.*, Spalek, K.*, Rasch, B., Gschwind, L., Coynel, D., Fastenrath, M., Papassotiropoulos, A., & de Quervain, D. J.-F. (2012). Psychoneuroendocrinology, 37(9), 1417-1424.
3. Genetic variants of NTRK2 are associated with emotion processing, a white-matter measure and DNA methylation levels in healthy young subjects.
Spalek, K., Coynel, D., Fastenrath,M., Freytag,V., Heck,A., Milnik,A., Vukojevic,V., de Quervain, D. J.-F., & Papassotiropoulos, A., in preparation (b).
4. PKC" is genetically linked to memory capacity in healthy subjects and to risk for posttraumatic stress disorder in genocide survivors.
de Quervain, D. J.-F., Kolassa, I.-T., Ackermann, S., Aerni, A., Boesiger, P., Demougin, P., Elberth, T., Ertl, V., Gschwind, L., Hadziselimovic, N., Hanser, E., Heck, A., Hieber, P., Huynh, K.-D., Klarhöfer, M., Luechinger, R., Rasch, B., Scheffler, K., Spalek, K., Stippich, C., Vogler, C., Vukojevic, V., Stetak, A., & Papassotiropoulos, A. (2012).
Proceedings of the National Academy of Sciences of the United States of America, 109(22), 8746-8751.
5. The association of the BDNF Val66Met polymorphism and the hippocampal volumes in healthy humans: A joint meta-analysis of published and new data.
Harrisberger, F.*, Spalek, K.*, Smieskova, R., Schmid, A., Coynel, D., Milnik, A., Fastenrath, M., Freytag, V., Gschwind, L., Walter, A., Vogel, T., Bendfeldt, K., de Quervain, D.J.-F., Papassotiropoulos, A., & Borgwardt, S. (2014). Neuroscience and Biobehavioral Reviews, 42C, 267-278.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
"!These authors contributed equally to this work.!
THEORETICAL BACKGROUND
11 2. THEORETICAL BACKGROUND
2.1. EMOTION
Throughout the history of science different definitions of emotion were formulated as a result of emotion theories. Historically the probably most influential theories were the James- Lange-Theory (1884) and the Cannon-Bard-Theory (1927). The James-Lange-Theory is based on the assumption that emotions are reactions to physiological changes in the body (Bear et al., 2009; Fehr & Stern, 1970). As a consequence of the criticism expressed on the James-Lange-Theory, the Cannon-Bard-Theory was developed (Bear et al., 2009; Cannon, 1987). This second theory proposes that after the emotional perception of a stimulus, the emotional experience follows and as a consequence emotions are expressed (Bear et al., 2009;
Cannon, 1987). The neurological basis of emotion, especially the role of the limbic system, was introduced by theories of James Papez in 1937 (Papez, 1995) and Paul MacLean in 1949 and 1952 (MacLean, 1955). The definition of the limbic system was and partially still is under great debate (for a review see LeDoux (2003) and Lewis, Haviland-Jones, & Feldman Barrett (2008)). Over time it became clear that it is very difficult to define an emotion theory applicable to the whole pallet of emotions and that it is very unlikely that only one brain system is responsible for all these different and complex processes. As a result one of the latest emotion theories formulated by LeDoux in the 90s, focuses only on the emotion of fear and discusses the involvement of different brain structures (LeDoux, 1998).
2.1.1. EMOTION PROCESSES
The term emotion in the every day understanding implicitly includes the differentiation of basically three distinct cognitive processes: perception, processing and response (for a review see Lang, Bradley, & Cuthbert (1998) and Lewis et al. (2008)). An emotional stimulus is first perceived, then processed and finally in most cases a response to the stimulus will follow. This differentiation is not scientifically determined and thus the terms are used sometimes interchangeably. In the context of experimental studies emotions are often induced visually by means of face photographs or natural scenes. The International Affective Picture System (IAPS; Lang, Öhmann, & Vaitl, 1988) is widely used and consists of a large set of standardized, emotion evoking, colour photographs covering a wide range of semantic categories. To investigate emotion perception, subjects are usually instructed to look at the presented stimuli, while emotion processing can be assessed from participant’s ratings after stimuli presentation. IAPS pictures are usually rated according to valence (ranging from
THEORETICAL BACKGROUND
pleasant to unpleasant) and arousal (ranging from calm to excited; Lang et al., 1998). An emotion response is assumed to engage processes from three reactive systems, namely (1) language expressions, (2) physiological changes and (3) behavioural reactions (Lang et al., 1998). These response types can be recorded for example through physiological measurements (e.g. heart rate, skin conductance) or behavioural assessments (observation and quantification of reactions).
2.1.2. NEURONAL BASIS OF EMOTION
Various brain structures have been found to be involved in emotional processes, but a key role is ascribed to the amygdala (Dalgleish, 2004). Most of the established knowledge on the topic, specifically in emotion perception and processing, comes from animal, human lesion and imaging studies. Several of these studies identified increased activation during emotion processing within a neuronal network of visual, limbic, temporal-parietal, prefrontal and subcortical areas including for example the amygdala, the insula, the medial prefrontal cortex (mPFC) and the ACC (for a review see Fusar-Poli et al. (2009); Phan et al. (2004) and Phillips, Drevets, Rauch, & Lane (2003)). Taylor, Phan, Decker, & Liberzon (2003) investigated activation differences between just passively viewing as opposed to viewing and rating IAPS pictures. They observed stronger activations in the amygdala and insula during passively viewing pictures, by contrast activation in medial frontal cortex (mFC) was only present during viewing and rating of the pictures. Although some studies investigated the underlying neuronal mechanisms of emotion processes by presenting faces to subjects and others by presenting IAPS pictures, there is evidence that these stimuli lead to similar brain activation patterns (Britton, Taylor, Sudheimer, & Liberzon, 2006; Hariri, Tessitore, Mattay, Fera, & Weinberger, 2002; Sabatinelli et al., 2011; see figure 1A). It nevertheless is noteworthy that in some regions faces seem to lead to stronger activations than IAPS pictures (Britton et al., 2006; Hariri et al., 2002). Furthermore, valence specific activation differences have been observed in limbic structures as well as insular and ventromedial prefrontal cortex (vmPFC; Britton et al., 2006; Fusar-Poli et al., 2009; see figure 1B).
THEORETICAL BACKGROUND
13 Figure 1. Brain activations during emotion perception and processing. A) Comparison of brain activation by emotional face (in blue) vs. emotional scene (in red) processing (in purple overlap of activations) from Sabatinelli et al. (2011). B) Valence-specific activation from the meta-analysis on face processing of Fusar-Poli et al. (2009).
2.2. MEMORY
What is usually understood under the term memory is the ability to encode, store and retrieve information and experiences. In scientific research, memory is understood as a very broad term incorporating many separate systems. Differentiation between these systems is based on temporal and content-related aspects as well as the involved brain structures and underlying molecular mechanisms.
2.2.1. MEMORY SYSTEMS
The temporal classification divides memory into short- and long-term memory systems (Bear et al., 2009; Squire, 1986). Short-term memory is further divided into immediate and working memory. Immediate memory encompasses the actively kept information since the beginning of information processing, which is in the focus of attention. Thus, its capacity is very limited (7 ± 2 items) and if information is not repeated, it is kept up for less than 30 seconds (Squire & Kandel, 2009). If contents from immediate memory are repeated and
THEORETICAL BACKGROUND
edited, thus actively kept in mind for several minutes this is referred to as working memory (Squire & Kandel, 2009). Some of the information coming either from short-term memory systems or directly as input, independently of short-term memory, reaches the long-term memory (Bear et al., 2009). Long-term memory encompasses memories lasting for hours, weeks, months or even a whole life-time and is divided into two main sub-systems based on the content and consciousness: the declarative or explicit memory and the non-declarative or implicit memory (for an overview see figure 2; Bear et al., 2009; Squire, 1986, 1987; Squire
& Zola, 1996; Squire, 2004).
Declarative memory is defined as the conscious recall of facts and events. More specifically, memory about facts, like the knowledge that the capital of Switzerland is Bern, is referred to as semantic memory and does not necessarily include the information about when and where this information was acquired (Bear et al., 2009; Squire, 1986; Squire & Zola, 1998; Squire, 2004). Whereas the memory about events, e.g. yesterday evening I had a nice dinner at the new restaurant in town, is called episodic memory and usually contains the information about where and when the event occurred (Bear et al., 2009; Squire, 1986; Squire
& Zola, 1998; Squire, 2004). In general, declarative memory consists of four processing stages: encoding, consolidation, recall and forgetting (Squire & Kandel, 2009). First the information is processed (encoding) and then it is saved (consolidation). If needed the information is reproduced (recall) and maybe with the course of time lost (forgetting).
The non-declarative memory is activated without our conscious awareness and is rather expressed through performance (Bear et al., 2009; Squire, 1987; Squire & Zola, 1996; Squire, 2004). It is further subdivided into categories like procedural memory (memory for skills e.g.
biking), priming (facilitation of stimuli identification due to previous exposure) as well as perceptual learning (ability to discriminate perceptual attributes due to previous exposure), simple classical conditioning (memory about a relationship of two stimuli), and non- associative learning including habituation (reduced reaction to a neutral stimulus because of repeated exposure), and sensitization (strong reaction to a otherwise neutral stimulus based on exposure to a previous aversive stimulus; Bear et al., 2009; Squire, 1987; Squire & Zola, 1996; Squire, 2004; Squire & Kandel, 2009). Non-declarative memory is less flexible than declarative memory in the sense that the acquired knowledge is not available to systems not involved in the initial learning (Squire & Zola, 1996).
THEORETICAL BACKGROUND
15 Figure 2. Overview of memory systems adapted from Squire (2004).
2.2.2. NEURONAL BASIS OF MEMORY
Several brain structures are involved in the functioning of the different memory systems (for an overview see figure 2). Immediate memory was observed to show a stimulus-specific activation, namely activations in occipital to temporal cortex (ventral stream) during the processing of form and quality of a stimulus and in occipital to parietal cortex (dorsal stream) when processing the location of a stimulus (Hautzel et al., 2002; Squire & Kandel, 2009).
Working memory involves in addition to this occipital-temporal-parietal network the prefrontal cortex (PFC; Bear et al., 2009; Squire & Kandel, 2009). Independent of the type of stimuli (e.g. verbal, object, spatial information) the same areas within the PFC seem to be involved in working memory (Hautzel et al., 2002). The PFC is suggested to exert a top-down control on the other brain regions activated by the processed information, in the sense of maintaining their activation (Gazzaley & Nobre, 2012; Squire & Kandel, 2009).
In long-term memory, according to the model of Squire (2004), the regions involved in the two types of declarative memory are supposed to be the same, whereas brain structures important for non-declarative memory are very heterogeneous. Central structures for semantic
MEMORY
SHORT-TERM LONG-TERM
IMMEDIATE WORKING DECLARATIVE
(EXPLICIT) NON-DECLARATIVE
(IMPLICIT)
SEMANTIC (FACTS) EPISODIC
(EVENTS) PROCEDURAL
(SKILLS, HABITS)
PRIMING AND PERCEPTUAL
LEARNING
ASSOCIATIVE LEARNING (SIMPLE CLASSICAL
CONDITIONING)
NON-ASSOCIATIVE LEARNING (HABITUATION, SENSITIZATION)
MEDIAL TEMPORAL LOBE DIENCEPHALON
NEO- CORTEX
STRIATUM AMYGDALA REFLEX
PATHWAY CEREBELLUM
SKELETAL RESPONSES EMOTIONAL
RESPONSES
PREFRONTAL CORTEX OCCIPITAL-PARIETAL/TEMPORAL
CORTEX
THEORETICAL BACKGROUND
and episodic memory are the medial temporal lobe (MTL) and the diencephalon (Bear et al., 2009; Squire, 2004; Squire & Wixted, 2011; Squire & Kandel, 2009). The medial temporal lobe encompasses different structures like the amygdala, the hippocampus and the surrounding cortex (parahippocampal, perirhinal and entorhinal cortex). From animal studies and studies on patients with lesions it emerged that especially the hippocampus and its surrounding cortex play an essential role in declarative memory (Bear et al., 2009; Squire, 1992; Squire & Wixted, 2011; Squire & Kandel, 2009). Furthermore, the diencephalon specifically the anterior and dorsal medial nuclei of the thalamus, mammillary bodies in the hypothalamus, and mammillo-thalamic tract, are central for functional declarative memory (Bear et al., 2009; Squire & Wixted, 2011). This is explainable by their connection to the MTL (Bear et al., 2009; Squire & Wixted, 2011). The most important output of the hippocampus is an axon bundle, so-called fornix. Most of its axons lead to the mammillary bodies, and its neurons project to the anterior thalamic nuclei. In general, the role of the MTL seems to be restricted time wise (Squire & Kandel, 2009). In the initial stage of memory formation several cortical structures as well as the MTL are involved. In the time after memory storage, the information gets reorganized and stabilized, not depending on the hippocampus any more. When this information is recalled the same cortical regions, which were processing the information, are reactivated depending gradually less on the MTL structures and more on the neocortex (Squire & Kandel, 2009).
On the other hand, non-declarative memory depends, based on the specific memory sub- system, on completely distinct brain structures. Procedural memory involves mainly a part of the basal ganglia, the striatum (nucleus caudatus and putamen; Bear et al., 2009; Squire, 1992;
Squire, 2004). The striatum receives information from the frontal and parietal cortex and its output is transferred to some thalamic nuclei as well as cortical regions involved in the motoric response (Bear et al., 2009). In priming and perceptual learning the neocortex is supposed to play a central role (Buckner & Koutstaal, 1998; Schacter & Buckner, 1998;
Squire, 1992, 2004). The specific neuronal location can vary with stimulus characteristics (Schacter & Buckner, 1998) and for perceptual learning is strongly task- as well as training- specific (Squire & Kandel, 2009). Usually decreased activations in regions involved in the prior processing of the stimulus are observed, pointing to a more efficient processing after exposure to the stimulus (Buckner & Koutstaal, 1998; Schacter & Buckner, 1998). The associative learning such as the classical conditioning, involves the amygdala for emotional responses and the cerebellum for skeletal responses, while the non-associative learning
THEORETICAL BACKGROUND
17 including habituation and sensitization is based on reflex pathways (Squire, 1992; Squire, 2004; Squire & Kandel, 2009).
2.2.3. MOLECULAR BASIS OF MEMORY
The investigation of molecular processes underlying short-term and long-term memory formation is mainly based on the work of Kandel and colleagues using the sea snail Aplysia Californica as a model organism (Kandel, 2001; Kandel, 2012; Squire & Kandel, 2009).
Specifically, the animals gill withdrawal reflex was used to model three different forms of implicit memory such as habituation, sensitization and conditioning (Kandel, 2001; Kandel, 2012). Whereas major differences in the molecular mechanisms appear between short-term and long-term memory, similar mechanisms are implied in implicit and explicit memory (Barco, Bailey, & Kandel, 2006; Kandel, 2001; Kandel, 2012). Above all, most of these molecular mechanisms investigated in the Aplysia seem to be conserved from invertebrates to mammals (Clapp, Hamm, Kirk, & Teyler, 2012; Kandel, 2001; Kandel, 2012).
Short-term memory consists in the modification of pre-existing proteins as well as synaptic connections (Kandel, 2001; Kandel, 2012; Mayford, Siegelbaum, & Kandel, 2012;
Squire & Kandel, 2009). In the short-term phase of sensitization, produced by a single tail shock, serotonin is released, binds to a serotonin receptor and activates a molecular signalling cascade. Frist, the adenylyl cyclase (AC) enzyme is activated and converts adenosine triphosphate (ATP) to the second messenger cyclic adenosine monophosphate (cAMP). This in turn, activates the cAMP-dependent protein kinase A (PKA) by binding to the regulatory subunits of PKA (see figure 3, spindles) and thus mobilizes the catalytic or active subunits (see figure 3, ovals). These active subunits increase neurotransmitter release by (1) closing the potassium (K+) channels hence increasing the calcium (Ca2+)-influx and by (2) directly acting on proteins in a Ca2+-independent manner involved in mobilization, fusion and release of neurotransmitter vesicles (for overview see figure 3). On the other hand, the long-term phase of sensitization implies synaptic changes such as activation of gene expression, protein synthesis and formation of new connections (Bailey & Kandel, 2008; Kandel, 2001; Kandel, 2012; Mayford et al., 2012; Squire & Kandel, 2009). In the case of repeated tail shocks leading to long-term sensitization, cAMP concentration stays elevated for a longer time period and the catalytic subunits of PKA activate mitogen-activated protein kinase (MAPK).
Both the catalytic subunits and MAPK are translocated to the nucleus, where PKA activates cAMP response element binding protein-1 (CREB-1) and MAPK deactivates the CREB-1 suppressor cAMP response element binding protein-2 (CREB-2). By binding to the cAMP
THEORETICAL BACKGROUND
response element (CRE) in the promoter of target genes, CREB-1 activates several genes including ubiquitin hydrolase. This results in persistent activity of PKA, and the transcription factor CCAAT-box-enhancer binding protein (C/EBP), which in combination with the activating factor (AF) activates downstream genes such as elongation factor 1" (EF1") leading to growth of new synaptic connections (for overview see figure 3; Kandel, 2001;
Kandel, 2012; Mayford et al., 2012; Squire & Kandel, 2009).
Figure 3. Molecular mechanisms of short- and long-term memory in A) the Aplysia during sensitization and in B) the hippocampus, specifically in the Schaeffer collateral, of mice (from Kandel (2001)).
Explicit memory such as implicit memory has a short-term phase not involving protein synthesis and a long-term phase requiring protein synthesis. In 1972, Bliss and Lomo discovered the concept of long-term potentiation (LTP; Bliss & Gardner-Medwin, 1973; Bliss
& Lomo, 1973), which is defined as activity-dependent plasticity resulting in a persistent enhancement of synaptic transmission and divided into early- and late-phase LTP (Bliss &
Gardner-Medwin, 1973; Bliss & Lomo, 1973; Malenka & Nicoll, 1999). Given the central role of the hippocampus in episodic memory (Squire & Wixted, 2011), the first LTP observation in the hippocampus (Bliss & Collingridge, 1993; Bliss & Gardner-Medwin, 1973;
Bliss & Lomo, 1973), and the inhibition of hippocampus-dependent memory by LTP- blockade using an N-Methyl-D-Aspartate (NMDA)-inhibitor (Ekstrom, Meltzer, McNaughton, & Barnes, 2001; Kandel, 2001; Shapiro, 2001), evidence was provided for the involvement of LTP and NMDA-receptors in explicit memory. The early-phase of LTP requires different signalling than the short-term phase of implicit memory (for a overview see figure 3; Barco et al., 2006). As depicted by example in figure 3B, in the Schaeffer collateral
THEORETICAL BACKGROUND
19 in the hippocampus, single high-frequent stimulation activates NMDA-receptors through binding of glutamate and removing the magnesium-ion, which is blocking the channel.
Hence, Ca2+-influx is enabled into the postsynaptic cell. The Ca2+ binds to Ca2+/calmodulin and activates three protein kinases: Ca2+-calmodulin-dependent kinase II (CaMK II), the protein kinase C (PKC) and the tyrosine kinase (not all are depicted in figure 3). CaMK II not only phosphorylates the NMDA-receptors, thus increasing their responsivity to glutamate, but also triggers the integration of new "-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)-receptors into the postsynaptic membrane (Lisman, Schulman, & Cline, 2002).
Since NMDA-receptors are not, as AMPA-receptors, active during routine synaptic transmission, the conductance change of AMPA-receptors and the integration of new AMPA- receptors could explain how the LTP is stabilized and maintained (Huganir & Nicoll, 2013;
Squire & Kandel, 2009). The late-phase of LTP is based on the same signalling pathway as the long-term phase of implicit memory. After repeated stimulation by action potentials the Ca2+-influx additionally activates AC and consequently activates the cAMP, PKA, MAPK and CREB signalling pathway, as well as synaptic growth. Furthermore, modulatory inputs like dopamine can modulate AC activation. Phosphatases such as calcineurin can influence the balance between protein phosphorylation-dephosphorylation and through this can constrain the late-phase of LTP and consequently memory (Malleret et al., 2001; Zeng et al., 2001). An additional mechanism contributing to LTP is an increase of neurotransmitter release in the presynaptic cell, which is suggested to be modulated by a retrograde mechanism from the postsynaptic cell as par example nitric oxide (NO; Hardingham, Dachtler, & Fox, 2013; Squire & Kandel, 2009). Importantly, the mechanism of late-phase LTP does not develop in any synapse, but only in synapses that were activated before and thus received a temporally limited synaptic tag, therefore referring to this process as synaptic tagging or capturing (Frey & Morris, 1997; Martin & Kosik, 2002; Redondo & Morris, 2011; Squire &
Kandel, 2009).
2.3. MEMORY MODULATION THROUGH EMOTION
It is well known that emotional information is better remembered than neutral one; a condition known as the emotional memory enhancement effect (McGaugh, 2003). Enhanced memory for emotional information is essential in evolutionary terms, since remembering dangerous and favourable situations is pivotal for survival and proliferation. Specifically the more arousing information is perceived, the more likely it will be remembered (LaBar &
THEORETICAL BACKGROUND
Cabeza, 2006). Furthermore, the emotional memory enhancement effect is partially mediated through noradrenergic activation of the basolateral amygdala (BLA; Cahill, Haier, Fallon, Alkire, Tang, Keator, Wu, & McGaugh, 1996; McGaugh, 2002, 2004), upon release of stress hormones like glucocorticoids and epinephrine (Roozendaal & McGaugh, 2011). Finally, the influence of BLA upon other brain structures like the hippocampus, caudate nucleus, nucleus accumbens and several cortical regions is crucial in the memory enhancement effect of emotion (for an overview see figure 4; McGaugh, 2002; Phelps, 2004; Roozendaal &
McGaugh, 2011).
Figure 4. Schematic representation of the components of the emotional memory enhancement effect (NE = norepinephrine; from Roozendaal & McGaugh (2011)).
There is evidence that the BLA modulates hippocampal LTP (Roozendaal & McGaugh, 2011). Specifically stimulation of BLA enhances LTP in the dentate gyrus of the hippocampus (Roozendaal & McGaugh, 2011). This reinforcement of hippocampal LTP is also embedded in the context of the emotional tagging, where it is suggested that an arousing emotional event activates a cascade of processes, hence leading to the supply of plasticity- related proteins to tagged synapses and thus can convert early-phase LTP to late-phase LTP (Bergado, Lucas, & Richter-Levin, 2011; McReynolds & McIntyre, 2012).
Several studies provide evidence for an influence of emotional arousal not just in the consolidation stage, but already in the encoding stage (for review see Hamann (2001)), which means, the emotional memory enhancement effect is observed already a few minutes after stimulus presentation as well as when emotional and neutral stimuli alternate rapidly (within
THEORETICAL BACKGROUND
21 seconds) underlining the emotional specificity of this effect (Canli, Zhao, Brewer, Gabrieli, &
Cahill, 2000; Hamann, Ely, Grafton, & Kilts, 1999; de Quervain et al., 2007).
2.4. MODULATING FACTORS OF EMOTIONAL PROCESSING AND EPISODIC MEMORY
Both processes which are in the focus of this thesis, namely emotional processing and episodic memory are very complex traits and thus can be additionally influenced by a broad spectrum of other modulating factors like hormones, sex, genetics, environmental factors as well as interactions between these factors. This thesis specifically has explored two of these factors being sex and genetics.
2.4.1. SEX
In what concerns emotional processing there is evidence that men and women react differently on emotional material (Gard & Kring, 2007). Particularly for aversive material, it was shown that women rate the emotional stimuli as more arousing in comparison to men and in addition react stronger to aversive pictures measured in physiological responses like event- related potentials (ERPs), electromyography (EMG), and startle response (Bradley et al., 2001; Gard & Kring, 2007; Lithari et al., 2010). These sex differences might in part be due to differences in hormonal levels of gonadal hormones, vasopressin, and oxytocin (Andreano &
Cahill, 2009; Ertman, Andreano, & Cahill, 2011; Honk & Schutter, 2007; Kret & Gelder, 2012; Meyer-Lindenberg, Domes, Kirsch, & Heinrichs, 2011; Uzefovsky, Shalev, Israel, Knafo, & Ebstein, 2012) as well as several environmental factors like gender-stereotypic socialization and socio-moral explanation (Fischer, Rodriguez Mosquera, Vianen, &
Manstead, 2004; Mathieson & Banerjee, 2011). These behavioural sex-dependent differences in emotion processing are as well visible at a neuronal level (for a review see Andreano &
Cahill (2009) and Stevens & Hamann (2012)). Specifically, stronger brain activations in women are observed during the processing of negative emotions, whereas men show increased activations when processing positive emotions (for an overview see figure 5;
Stevens & Hamann, 2012).
Regarding episodic memory performance there is evidence that females outperform males (Andreano & Cahill, 2009; Bloise & Johnson, 2007; Herlitz, Nilsson, & Bäckman, 1997; Herlitz, Reuterskiöld, Lovén, Thilers, & Rehnman, 2013; de Frias, Nilsson, & Herlitz,
THEORETICAL BACKGROUND
2006). This female’s advantage can already be observed in childhood and puberty (Herlitz et al., 2013; Kramer, Delis, Kaplan, O’Donnell, & Prifitera, 1997) and is stable over time in adulthood and older age (de Frias, Nilsson, & Herlitz, 2006). In a recent study of (Young, Bellgowan, Bodurka, & Drevets, 2013) higher activations in dorsolateral prefrontal cortex (DLPFC), dorsal anterior insula, and precuneus were observed in women compared to men during the recall of autobiographic memories.
Figure 5. Sex-dependent brain activation differences during emotion processing A) of negative emotions and B) of positive emotions (from Stevens & Hamann (2012)).
There is evidence that also for emotional memory women show enhanced performance compared to men (Andreano & Cahill, 2009). Several studies observe sex-dependent brain activation differences specifically in the amygdala in subsequent emotional memory contrasts (Andreano & Cahill, 2009; Cahill, 2003; Hamann, 2005). There is however a lack of studies investigating whole-brain wide sex-dependent activation differences in emotional memory. In general, it has to be considered that the evidence for better performance of females compared to males is mostly based on episodic memory recall tasks of life events (Andreano & Cahill 2009).
Taken together, it is unclear so far if the behavioural differences between the sexes in emotional processing and episodic memory, specifically emotional memory, are two independent processes or linked to each other. More specifically, female’s stronger perception of emotionally arousing information could lead to stronger encoding thereby inducing an advantage in episodic emotional memory performance. In the publication “Gender-dependent dissociation between emotional appraisal and memory: A large–scale behavioural and fMRI study” (Spalek et al., in preparation (a)) we investigated the relationship between sex-
THEORETICAL BACKGROUND
23 dependent differences in emotional processing and emotional memory performance as well as the underlying neuronal patterns. Additionally, it is known that testosterone can exert its influence through binding to androgen and estradiol receptors in brain regions like the amygdala and the hippocampus, which are involved in emotional processing as well as emotional memory (Abdelgadir, Roselli, Choate, & Resko, 1999; Beyenburg et al., 2000;
Kritzer, 2004; Roselli, Klosterman, & Resko, 2001; Sarkey, Azcoitia, Garcia-Segura, Garcia- Ovejero, & DonCarlos, 2008; Sarrieau et al., 1990; Simerly, Chang, Muramatsu, & Swanson, 1990). Several studies found an association between testosterone and affective behaviour (for a review see van Wingen, Ossewaarde, Bäckström, Hermans, & Fernández (2011)), as well as amygdala activation in response to biologically salient stimuli (Derntl et al., 2009; Hermans, Ramsey, & van Honk, 2008; Stanton, Wirth, Waugh, & Schultheiss, 2009) and memory performance (Barrett-Connor, Goodman-Gruen, & Patay, 1999; Cherrier et al., 2005; Cherrier et al., 2002; Fonda, Bertrand, O’Donnell, Longcope, & McKinlay, 2005; Hogervorst, Matthews, & Brayne, 2010; Moffat et al., 2002; Perry et al., 2001; Wolf & Kirschbaum, 2002;
Yonker, Eriksson, Nilsson, & Herlitz, 2006; Young, Neiss, Samuels, Roselli, & Janowsky, 2010). But it is not clear if testosterone might have an impact on memory perfromance by modulating amygdala reactivity during processing. Thus we aimed at analysing this possibility in our publication “Testosterone levels in healthy men are related to amygdala reactivity and memory performance” (Ackermann et al., 2012).
2.4.2. GENETICS
Emotion (Bevilacqua & Goldman, 2011) and episodic memory (Papassotiropoulos & de Quervain, 2011) as well as emotional episodic memory (Todd, Palombo, Levine, &
Anderson, 2011) are complex polygenetic behavioural traits, influenced by genetic and environmental factors as well as by gene-environment interactions. These traits have substantial heritability estimates varying between 30% - 60% (Bevilacqua & Goldman, 2011;
Papassotiropoulos & de Quervain, 2011). Given the broadness of the genetically driven influence on these cognitive traits, the focus of this thesis was on the role of three specific genes described in the following.
First, the NTRK2 gene, also known as tyrosine kinase receptor B (TRKB), is associated in several studies with various psychiatric disorders like depression, schizophrenia, addiction, eating and anxiety disorders (Alonso et al., 2008; Boulle et al., 2012; Deo et al., 2013; Ernst et al., 2011; Gupta, You, Gupta, Klistorner, & Graham, 2013; Hauger, Risbrough, Oakley, Olivares-Reyes, & Dautzenberg, 2009; Hill, 2012; Kohli et al., 2010; Mahan & Ressler, 2012;
THEORETICAL BACKGROUND
Marsden, 2013). Given that the dysregulation of emotional processes is a common characteristic of many psychiatric disorders (Cole, Michel, & Teti, 1994; Kring & Sloan, 2009) and the evidence for the involvement of NTRK2 in these disorders, it can be hypothesized that NTRK2 is genetically associated with emotion processing. Since there is a lack of studies examining the role of NTRK2 in emotional processing in healthy subjects, we investigated this aspect in the publication “Genetic variants of NTRK2 are associated with emotion processing, a white-matter measure and DNA methylation levels in healthy young subjects” (Spalek et al., in preparation (b)).
Second, following the considerable evidence for an important role of protein kinases like PKA, PKC, CaMK II and MAPK in memory formation from animal and human studies (Sun
& Alkon, 2014; Xu, Liu, & Alkon, 2014; de Quervain & Papassotiropoulos, 2006), as well as emotional memory formation based only on animal studies (McGaugh, 2000; Rodrigues, Schafe, & LeDoux, 2004), we investigated the role of genes encoding for protein kinases in human emotional memory in the publication “PKC" is genetically linked to memory capacity in healthy subjects and to risk for posttraumatic stress disorder in genocide survivors“ (de Quervain et al., 2012).
Finally, BDNF, encoded by the BDNF gene, is suggested to be involved in synaptic plasticity (Bliss & Cooke, 2011; Lu, Christian, & Lu, 2008; Martin & Kosik, 2002). BDNF is highly expressed in the hippocampus (Binder & Scharfman, 2004; Wetmore, Ernfors, Persson, & Olson, 1990), a central brain structure for learning and memory (Squire & Wixted, 2011). Additionally, BDNF has been shown to play a role in learning and memory processes (Baj, Carlino, Gardossi, & Tongiorgi, 2013; Cunha, Brambilla, & Thomas, 2010). Thus, many studies investigated the association of BDNF gene with hippocampal volumes, but so far results are very inconsistent. In the publication “The association of the BDNF Val66Met polymorphism and the hippocampal volumes in healthy humans: A joint meta-analysis of published and new data“ (Harrisberger et al., 2014) we investigated the association of a genetic variant of the BDNF gene (rs6265) and hippocampal volumes. With the aim to increase the statistical power of our results, we conducted a meta-analysis and combined it with our own study data. We additionally addressed the influence of potential moderators such as measuring technique, magnetic field strength, age, gender, ethnicity, Val/Met ratio, sample size, quality rating, hippocampal volumes normalized to intracranial volume (ICV), and publication year.
METHODS
25 3. METHODS
3.1. NEUROIMAGING
In the last years, imaging techniques have found extensive applications in neuroscience research. Imaging techniques based on magnetic resonance imaging (MRI) like functional magnetic resonance imaging (fMRI), diffusion weighted imaging (DWI), resting state and structural measures are very commonly used ones next to positron emission tomography (PET). The results of this thesis are based on fMRI, diffusion tensor imaging (DTI) and structural measurements.
fMRI is used for measuring neuronal activity by recording changes in cerebral blood flow. The concept of fMRI is founded upon the blood oxygenation level-dependent (BOLD) response, which basically states that the measured MR signal changes in dependence of the ratio of oxygenated vs. deoxygenated blood, given the different magnetic properties the blood has in these two states. It is assumed that oxygenated blood is delivered to activated brain regions in order to enable the increased neuronal activity, in form of the so-called hemodynamic response function (HRF). The HRF can vary across subjects, between brain regions and between tasks (Waugh, Hamilton, & Gotlib, 2010). Compared to other imaging techniques fMRI has many advantages such as that it is non-invasive, it has a relatively short acquisition time, and good spatial as well as reasonable temporal resolution. Despite the often mentioned limitation that the BOLD response is an indirect measure, a study of Logothetis, Pauls, Augath, Trinath, & Oeltermann (2001) provides evidence for the BOLD signal reflecting the input and processing of neuronal information in a specific area. Other limitations of fMRI are that there is no distinction between excitatory and inhibitory connections, further the signal might be influenced by large vessels (even when they are located far from the activation site), and it has to be considered that one brain voxel contains many different physiological components (e.g. around five million neurons, 2e10 to 5e10 synapses, around 220 km of axons; Logothetis, 2008).
DTI is a subcategory of DWI and is usually used to measure the diffusion of water as a function of spatial location (Johansen-Berg & Behrens, 2009). The functional principle that enables DTI is that water molecules tend to diffuse more freely along the axon fibres, thus the measure of water diffusion relates to axonal orientation (Johansen-Berg & Behrens, 2009). In gray matter water diffusion is largely independent of tissue orientation (isotropic), whereas in white matter diffusion is mostly tissue orientation-dependent (anisotropic; Johansen-Berg &
Behrens, 2009). Two commonly used measures in DTI are (1) fractional anisotropy (FA), which is a measure of the directional dependence of diffusion (Basser, 1995) and reflects fibre
METHODS
density as well as coherence within a voxel (Beaulieu, 2002), and (2) mean diffusivity (MD), which reflects the magnitude of water diffusion within a voxel and depends on the density of physical obstructions like membranes and the distribution of water molecules between different cellular compartments (Beaulieu, 2002; Sen & Basser, 2005). There are some limitations of DTI, which have to be considered like the inferential character of resulting white matter properties, possible failure of connection identification, further displayed white matter connections do not have to be functional (representation of anatomy not function), and its sensitivity to artefacts introduced through reduced spatial resolution (usually due to the commonly applied echo-planar imaging technique (EPI)), subject motion and periodic ventricular pulsations with each heart beat (Alger, 2012).
Structural measures can be used to assess brain volume differences among subjects (for a review see Caviness, Lange, Makris, Herbert, & Kennedy (1999)). For the analysis volumetric data are segmented into cortical and subcortical structures as well as cerebrospinal fluid (CSF). Segmentation can be done either manually or automatically. Even though manual segmentation is generally considered as the gold standard due to the precise delineation of anatomical structures, the increasing sample size in imaging studies renders the process of manual segmentation less practicable, as it is both costly and time consuming. For automated segmentation there are different tools available like FreeSurfer or fMRI of the brain software library (FSL).
Importantly, in all these three types of imaging measurements the brain is divided into thousands of voxels, which makes a correction for multiple testing necessary in analysis.
Additionally, if not main interest of the imaging analysis controlling for sex, age and in the case of structural data for ICV will allow to address their potential influence.
3.2. PSYCHONEUROENDOCRINOLOGY
The field of psychoneuroendocrinology incorporates the investigation of the association between psychological concepts with neuronal correlates and the endocrine system. To examine the role of the endocrine system in this interplay different approaches are used such as measuring hormone levels from blood or saliva as well as pharmacological manipulation by administering the hormone being investigated. When performing statistical analyses of endocrine measures, it is important to take into account possible moderating factors like circadian rhythmicity, sex, as well as age. Research applying the psychoneuroendocrinological approach found a wide application.
METHODS
27 Numerous studies focused specifically on the psychoneuroendocrinological aspects of emotional processing, memory as well as episodic memory, investigating for instance the role of cortisol (Ackermann, Hartmann, Papassotiropoulos, de Quervain, & Rasch, 2013; Het, Ramlow, & Wolf, 2005; Wolf, Kuhlmann, Buss, Hellhammer, & Kirschbaum, 2004; van Ast et al., 2013), estrogen (Gasbarri, Tavares, Rodrigues, Tomaz, & Pompili, 2012; Pompili, Arnone, & Gasbarri, 2012) and testosterone (Thilers, Macdonald, & Herlitz, 2006; van Wingen et al., 2011).
3.3. GENETIC ANALYSIS
In general, when performing genetic analyses it is crucial that the investigated trait has substantial heritability rates. Given this prerequisite basically two different genetic approaches can be used to analyse the data, namely linkage and association studies (Papassotiropoulos & de Quervain, 2011). Linkage studies are usually performed in pedigrees in the context of a disease that represents the trait of interest. They aim at identifying a genetic variation, at an unknown trait locus, which is associated with the disease, by taking advantage of the linkage between the trait locus and a marker with known location. If the trait locus is in linkage disequilibrium with the marker locus, then the alleles of both loci are very likely to be inherited together (cosegregation during meiosis). Various methods can be used to identify the trait-related genetic variant like positional cloning, fine-mapping and in-depth resequencing (Neale, Ferreira, Medland, & Posthuma, 2008; Papassotiropoulos & de Quervain, 2011). On the other hand, association studies compare the genotype frequencies of common genetic polymorphisms between groups of unrelated (case-control design) or related (family-based design) individuals, and are well suited for application to quantitative traits like for instance episodic memory (Neale et al., 2008; Papassotiropoulos & de Quervain, 2011).
Within the class of association studies further two distinct study types are distinguished, namely candidate gene studies and GWAS. Candidate gene studies are usually applied when there is prior knowledge about the gene’s (or several genes) biological relevance for the trait or disease of interest. This approach is hypothesis driven and very focused, but bias afflicted and prevents the identification of novel genes. Thus, if the aim is to identify a novel gene (or several genes), which is (are) associated with a trait or a disease, GWAS are the method of choice. In a GWAS analysis, millions of genetic variants located at different position on the entire genome are tested for an association with the trait or disease of interest (Papassotiropoulos & de Quervain, 2011). Although all these approaches are extensively used, a considerable amount of the heritability of complex phenotypes like episodic memory
METHODS
or emotion is still unexplained, and therefore has been termed as “missing heritability”
(Papassotiropoulos & de Quervain, 2011). In an effort to improve this unsatisfactory situation, approaches focusing on gene-gene interactions, gene set analysis (GSA) and genetic networks are becoming more popular and elaborated. Analysis addressing gene-gene interactions, for example in the sense of statistical epistasis, are used to investigate interactions between loci of genes (for a review see Cordell (2009)). GSA like for instance pathway-based analysis (PBA) goes a step further by examining the association of a set of genes, which are biologically related in the sense of having similar function (thus being summarized as a pathway), with the given disease or trait (for a review see Holmans (2010) and Wang, Li, &
Hakonarson (2010)). Genetic analysis applying a network approach like network-assisted analysis (NAA) investigates as well the association between gene sets and phenotype, but is not using pre-defined gene sets like in the case of PBA, instead defines gene sets based on sub-network search algorithms (for a review see Jia & Zhao (2014)).
3.4. EPIGENETIC ANALYSIS
The field of epigenetics investigates processes in cells, which take place above the level of genetics, namely processes that alter gene function without changing the deoxyribonucleic acid (DNA) sequence (Fazzari & Greally, 2010; Sweatt, 2013). In the recent years epigenetics have received a lot of attention in research and a broad range of applications in several domains. Within this thesis the focus is restricted to epigenetics in neuroscience, also termed as neuroepigenetics by Sweatt (2013). In his review Sweatt (2013) divides the major epigenetic molecular mechanisms into eight different categories. DNA cytosine methylation, a subcategory of the covalent DNA modification category, is one of the most studied epigenetic mechanisms given its strong regulatory influence on gene transcription (Fazzari &
Greally, 2010; Sweatt, 2013) and is within the focus of this thesis. DNA methylation seems to occur preferentially at cytosine-guanine dinucleotide (CpG) DNA sequences, so called CpG sites, but according to recent findings can occur as well at non-CpG sites (Fazzari & Greally, 2010; Sweatt, 2013). Various statistical methods, already used in analysis of other large data sets, are applied as well in methylation analysis (for a review see Fazzari & Greally (2010)).
Methodological approaches for statistical analyses of methylation data are still emerging.
Different normalization, pre- and post-processing strategies for large-scale methylation data (in the case of whole-genome methylation data) are getting introduced. For a methodological approach proposal see Milnik et al. (in preparation).
ORIGINAL RESEARCH PAPERS
29 4. ORIGINAL RESEARCH PAPERS
4.1. GENDER-DEPENDENT DISSOCIATION BETWEEN EMOTIONAL APPRAISAL AND MEMORY: A LARGE–SCALE BEHAVIOURAL AND FMRI STUDY
Spalek, K., Fastenrath, M., Ackermann, S., Auschra, B., Coynel, D., Frey, J., Gschwind, L., Hartmann, F., van der Maarel, N., Papassotiropoulos, A., de Quervain, D. J.-F., & Milnik, A., in preparation (a).
#$%&'(!)'*+',-+'.'*+'*%!+$//01$2%$0*!3'%4''*!'50%$0*2&!2..,2$/2&!2*+!5'50,6(!7!&2,8'9/12&'!
"!
3':2;$0<,2&!2*+!=>?@!/%<+6A!
B!
C! !
733,';$2%'+!%$%&'(!)'*+',!/%<+6!'50%$0*2&!2..,2$/2&!2*+!5'50,6!
D!
E! !
7<%:0,F/!*25'/!
G!
H.2&'I!JA"K!L2/%'*,2%:!>A"K!71I',52**K!HABK!7</1:,2K!MABK!N06*'&K!OA"K!L,'6!PA"K!)/1:4$*+K!QA"K!
R!
S2,%52**K!LABK!;2*!+',!>22,'&!TA"K!U2.2//0%$,0.0<&0/K!7ABKCKDK!+'!V<',;2$*!OA!PA-LA"KD!2*+!>$&*$I!7ABKD! W!
X! !
7==$&$2%$0*/!
"Y!
"!O$;$/$0*!0=!N08*$%$;'!T'<,0/1$'*1'K!O'.2,%5'*%!0=!U/61:0&086K!Z*$;',/$%6!0=!M2/'&K!DYYX!M2/'&K!
""!
H4$%[',&2*+!
"B!
B!O$;$/$0*!0=!>0&'1<&2,!T'<,0/1$'*1'K!O'.2,%5'*%!0=!U/61:0&086K!Z*$;',/$%6!0=!M2/'&K!DYYX!
"C!
M2/'&K!H4$%[',&2*+!
"D!
C!Q$='!H1$'*1'/!#,2$*$*8!L21$&$%6K!O'.2,%5'*%!M$0['*%,<5K!Z*$;',/$%6!0=!M2/'&K!DYYX!M2/'&K!
"E!
H4$%[',&2*+!
"G!
D!Z*$;',/$%6!U/61:$2%,$1!N&$*$1/K!Z*$;',/$%6!0=!M2/'&K!DYYX!M2/'&K!H4$%[',&2*+!
"R!
"W! !
N0,,'/.0*+$*8!2<%:0,!
"X!
>$&*$IK!7**'%%'!
BY!
Z*$;',/$%6!0=!M2/'&K!O$;$/$0*!0=!>0&'1<&2,!T'<,0/1$'*1'!
B"!
M$,52**/82//'!WK!NS-DYYX!M2/'&!!
BB!
U:0*'(!\D"!G"!BGR!YBGR!
BC!
2**'%%'A5$&*$I]<*$32/A1:!
BD!
BE! !
T<53',!0=!.28'/(!
BG!
T<53',!0=!=$8<,'/(!E^!T<53',!0=!%23&'(!C!
BR!
T<53',!0=!40,+/(!73/%,21%!^!@*%,0+<1%$0*!^!O$/1<//$0*!
BW!
! "!
#$%&'()*+,-*&./0!
1!
23*!45.3'6/!+*$)46*!&'!$'-7*.8&,!98&4&$84)!8&.*6*/./:!
"!
;*<('6+/0!/*=!+899*6*&$*/>!-*-'6<>!*-'.8'&4)!-*-'6<>!?4)*&$*>!46'5/4)>!78$.56*!.4/%>!('6+!
@!
.4/%! !
A!
